DeepSeek V4-Pro 75% Price Cut Permanent — AI Price War Started

Sponsored by

The AI price war just officially started. DeepSeek made its 75% V4-Pro price cut permanent — now $0.87 per million output tokens, a fraction of OpenAI, Anthropic, and Google pricing for comparable models.

Simultaneously, Anthropic publicly accused DeepSeek of "distillation attacks" — training V4-Pro on competitor model outputs. The price war just got messy AND diplomatic.

The World's Biggest Dev Event Hits Silicon Valley

WeAreDevelopers World Congress comes to San José, CA — September 23–25, 2026. 10,000+ developers, 500+ speakers, and the full software development lifecycle under one roof, in the heart of Silicon Valley.

Kelsey Hightower. Thomas Dohmke (fmr. CEO, GitHub). Christine Yen (CEO, Honeycomb). Mathias Biilmann (CEO, Netlify). Olivier Pomel (CEO, Datadog). The people actually building the tools you use every day — all on one stage.

AI, cloud, DevOps, security, architecture, and everything real builders ship with. Workshops, masterclasses, and the official congress party.

Tickets from $599. Use code GITPUSH26 for 10% off.

Secure Your Pass

The new pricing — per 1M tokens

DeepSeek's Chinese inference infrastructure runs at a fundamentally lower cost basis.

• DeepSeek V4-Pro: $0.27 in / $0.87 out (10× cheaper than Sonnet 4.6)

• GPT-4o mini: $0.15 in / $0.60 out (still cheaper, less capable)

• Claude Haiku 3.5: $1.00 in / $5.00 out

• Gemini 3.5 Flash: $1.50 in / $9.00 out

• Claude Sonnet 4.6: $3.00 in / $15.00 out

The 5 facts that matter

The AI price floor just shifted permanently lower.

1. Cut is permanent. DeepSeek specifically used the word "permanent" — not a promo.

2. Apache 2.0. You can self-host V4-Pro weights and pay $0 inference if you have GPUs.

3. Anthropic distillation accusations. Linguistic patterns in V4-Pro reportedly look Claude-derived. Unproven but it shifts the narrative.

4. Anthropic + Google pressured most. Their flagship pricing is now 10× DeepSeek's. Cuts likely Q3-Q4.

5. OpenAI's IPO timing hates this. A price war right before going public is the worst possible market story.

What to do this week

• Pure-text agentic workloads: Switch to DeepSeek V4-Pro via OpenRouter

• Multimodal: Stay on Gemini 3.5 Flash (DeepSeek is text-only)

• RAG with caching: DeepSeek's cache pricing is even better — stay there

• Regulated industries: Stay on US providers OR self-host DeepSeek weights for cost-without-data-risk

• Indie hackers: Test one workload. If quality holds, drop your AI bill 90%.

→ Read the full breakdown + comparison table + FAQ

📩 AI pricing wars + every major model launch. Free.

Subscribe here →

Sources: Semafor, ThenextWeb, OpenSource For U, Techzine Global, TipRanks, Reuters. — Tech4SSD Editorial